Incorporation of non-anticipativity constraint in stochastic optimization model for electricity contracts portfolio

When dealing with decision-making problems under great uncertainty and risk, a common situation in the economic-financial field, stochastic optimization models are widely used to minimize risk exposure or maximize profit. This kind of problem is also recurrent in the energy trading field, where fluctuations in prices and availability create unstable decision-making conditions. In this context, many optimization models evaluate decisions based on the implied use of future information, thereby generating unrealistic scenarios. In this regard, the article proposes using K-means-based grouping of similar PLD (Settlement Price of Differences) scenarios and the Non-Anticipativity Constraint to improve the results of optimization models. As a validation methodology, a direct replication and comparison with the model developed by Maia et al. [2023] was conducted. The major findings of this study include improved outcome stabilization and the measurement of the clairvoyance cost, which translates into a lower but actually achievable profit.